namba-ai: Skill for Claude Code

State effect: mutating workflow entry point. Use help/probe paths read-only, and otherwise expect repository state or GitHub state to change.

Generated instruction contract for this command skill:

• Purpose: keep the role or command scope explicit, bounded, and testable.
• Boundary: honor read-only versus mutating state effects, configured sandbox mode, and assigned file or workflow ownership.
• Required output: report concrete actions, changed paths or artifacts, validation evidence, and pass/fail status or blockers.
• Pass/fail criteria: claim success only when acceptance criteria and configured validation are satisfied; otherwise name the exact blocker and impact.
• Evidence expectations: cite source artifacts such as SPEC files, .namba/ configs, diffs, test output, PR/check links, or generated manifests instead of relying on unsupported assertions.
• Security responsibilities: never expose or commit secrets; treat auth, privacy, destructive commands, permission changes, and external network or credential use as security-sensitive.
• Destructive command and escalation policy: do not run destructive commands unless explicitly requested; request approval for privileged, networked, or sandbox-blocked actions only when the active approval mode allows it, and otherwise report the blocker or use a safe non-escalating path.
• Fallback implementer boundary: if a specialist path is unavailable and the main/default implementer takes over, stay within the assigned scope and preserve the same evidence and validation duties.
• Portability: keep durable guidance non-project-specific unless the current repository config or SPEC explicitly provides the project detail.

Use this skill when the user explicitly says $namba-plan, namba plan, or asks to create a new feature SPEC package.

Behavior:

• Before reading project docs, checking Git state, or running the CLI, run the Namba clarification gate on the user's raw request.
• If the request is short, broad, or missing target surface, user flow, scope boundaries, constraints, acceptance criteria, or validation, do not run namba plan yet. Ask 1-3 concise questions first; 게시판 만들어줘 is the canonical example that must ask questions before planning.
• Treat Codex Plan mode as the preferred clarification surface and namba plan as the executor: when native Plan mode choice UI is available, use it to ask the clarification questions before any CLI command.
• The Plan mode output must be converted into a refined description shaped as Goal, Scope, Constraints, and Acceptance, then passed to namba plan "<refined description>".
• Repo hooks and skills cannot force Codex to switch modes by themselves; if native Plan mode UI is unavailable in the current session, fall back to concise text questions in chat instead of inventing a CLI wizard.
• Continue the clarification loop across turns until you can restate the request as Goal, Scope, Constraints, and Acceptance. Only then invoke the CLI with the refined description.
• Prefer the installed namba plan CLI when available.
• Keep namba plan for feature-oriented SPEC work; use namba harness when the request is about reusable agent, skill, workflow, or orchestration scaffolding; use $namba-create when the user wants the repo-local skill or custom-agent artifact itself instead of another SPEC.
• When repo-managed MCP presets are configured, prefer them for planning context before broader web search; for example, use context7 for library and framework docs, sequential-thinking for deeper decomposition, and playwright for browser-verified flows.
• Read .namba/project/product.md, .namba/project/tech.md, .namba/project/mismatch-report.md, .namba/project/quality-report.md, and any relevant .namba/project/systems/*.md artifacts before drafting the SPEC.
• Treat executable code and authoritative config as stronger planning evidence than docs, and preserve code-vs-doc conflicts instead of smoothing them out.
• Keep planning in the current workspace. When branch-per-work is enabled, create or switch to the dedicated spec/... branch before writing .namba/specs/<SPEC>/.
• Use --current-workspace only when the user intentionally wants to scaffold on the current branch without creating a dedicated SPEC branch.
• Do not create a planning worktree. Reserve temporary worktrees for overlapping namba run SPEC-XXX --parallel execution, and expect them to disappear after the run finishes cleanly.
• Create the next sequential SPEC-XXX package under .namba/specs/ after that branch decision is explicit.
• Seed .namba/specs/<SPEC>/reviews/ with product, engineering, design, and aggregate readiness artifacts.
• Unless the invocation includes --no-review, treat the created SPEC as an automatic handoff and immediately continue with $namba-plan-review SPEC-XXX after the CLI prints the new SPEC ID.
• After the automatic handoff, read .namba/specs/<SPEC>/reviews/readiness.md; if it does not say Cleared reviews: 3/3, do not treat planning as complete. Continue with $namba-plan-review SPEC-XXX again, or run the missing $namba-plan-pm-review, $namba-plan-eng-review, and $namba-plan-design-review tracks before implementation.
• When --no-review is present, stop after scaffold creation and point any later review work to $namba-plan-pm-review, $namba-plan-eng-review, and $namba-plan-design-review, or $namba-plan-review SPEC-XXX.
• Keep the scope concrete and implementation-ready.